breadclipd: watch-driven capture, flock singleton, ignore rules, pin verb

- Capture now keys off wl-paste's `CLIPBOARD_STATE`: `sensitive`
  (password manager via `wl-copy --sensitive`, covers the old
  x-kde-passwordManagerHint case) and `nil`/`clear` persist nothing;
  `data` reads the content off stdin in the same event, so the
  sensitive check and the read can't straddle a clipboard change.
  A manual `--capture-once` with no CLIPBOARD_STATE falls back to
  querying wl-paste directly, now wrapped in `timeout -k 2 5` so a
  stalled offer can't block a capture process forever.
- Single-instance guard moves to `bread_utils::singleton` (flock) —
  kernel-atomic, auto-released on death, no stale pid file.
- `ignore_rules::is_sensitive`: conservative best-effort heuristics
  that skip copies which look like secrets even when unflagged —
  PEM/OpenSSH private key blocks, `password:`-style lines, labelled
  one-time codes, Luhn-valid card numbers, well-known token prefixes.
  A convenience, not a security boundary.
- `bread.command.clip.pin` verb (payload `{id, pin?}`, pin defaults
  true) → `set_pinned`, emits `bread.clip.pinned` / `.pin.failed`.
- Two independent `wl-paste --watch` loops (regular clipboard always
  on; primary selection when `capture.primary`), each restarting with
  2s→30s capped backoff and state-change-only journal logging.
- Image type is sniffed from magic bytes (PNG/JPEG) on the stdin path
  and requested by actual offered type on the fallback path — a JPEG
  is stored as `.jpg` / `image/jpeg`, not relabelled PNG.
- content_kind: a single line of prose that merely contains a keyword
  ("class is a concept") no longer classifies as code — it must look
  like a statement.

breadclipd/src/content_kind.rs

@@ -105,12 +105,29 @@ fn looks_like_code(text: &str) -> bool {
         })
         .count();
 
-    // Multi-line with a meaningful fraction of "code-shaped" lines, or an
-    // unambiguous single-line token (import/#include/fn signature), counts
-    // as code. A single short line of prose won't hit either bar.
-    token_hits >= 2
-        || (lines.len() > 1 && brace_or_semicolon_lines * 2 >= lines.len())
-        || (lines.len() == 1 && token_hits >= 1 && text.len() < 200)
+    // Multi-line with a meaningful fraction of "code-shaped" lines, or a
+    // couple of code tokens anywhere, counts as code.
+    if lines.len() > 1 {
+        return token_hits >= 2 || brace_or_semicolon_lines * 2 >= lines.len();
+    }
+
+    // A single line only counts as code if it actually *looks* like a code
+    // statement — a prose sentence that merely contains a keyword ("let me
+    // show you", "function of time", "class is a concept") must stay plain.
+    if lines.len() == 1 && token_hits >= 1 && text.len() < 200 {
+        let trimmed = text.trim_start();
+        let import_like = trimmed.starts_with("import ") || trimmed.starts_with("#include");
+        let code_shaped = trimmed.ends_with('{')
+            || trimmed.ends_with('}')
+            || trimmed.ends_with(';')
+            || trimmed.ends_with('(')
+            || trimmed.ends_with(')')
+            || trimmed.contains("=>")
+            || (trimmed.contains('(') && trimmed.contains(')'))
+            || trimmed.contains(" = ");
+        return import_like || code_shaped;
+    }
+    false
 }
 
 #[cfg(test)]
@@ -170,6 +187,15 @@ mod tests {
         assert_eq!(detect("import numpy as np"), "code");
     }
 
+    #[test]
+    fn single_line_prose_containing_a_keyword_is_plain() {
+        assert_eq!(detect("let me show you something"), "plain");
+        assert_eq!(detect("function of time"), "plain");
+        assert_eq!(detect("class is a concept"), "plain");
+        assert_eq!(detect("const means constant"), "plain");
+        assert_eq!(detect("def is short for define"), "plain");
+    }
+
     #[test]
     fn plain_prose_is_plain() {
         assert_eq!(

breadclipd/src/ignore_rules.rs

@@ -0,0 +1,182 @@
+//! Best-effort content heuristics for never persisting obvious secrets.
+//!
+//! The password-manager path (`CLIPBOARD_STATE=sensitive` / the
+//! `x-kde-passwordManagerHint` MIME type) only catches copies an app
+//! deliberately flagged. These rules catch copies that *look* like secrets
+//! even when the app didn't flag them — one-time codes, credit card
+//! numbers, private keys, credential lines. They are deliberately
+//! conservative (a false "skip" is cheaper than a leaked password) and they
+//! are a convenience, not a security boundary: the 0600 file permissions
+//! and 0700 data dir are the real protection.
+
+/// Would persisting this text be a bad idea? Skipped copies never reach the
+/// database (or the event bus), like the password-manager path.
+pub fn is_sensitive(text: &str) -> bool {
+    let trimmed = text.trim();
+    if trimmed.is_empty() {
+        return false;
+    }
+    looks_like_private_key(trimmed)
+        || looks_like_credential_line(trimmed)
+        || looks_like_otp(trimmed)
+        || contains_card_number(trimmed)
+        || contains_api_token(trimmed)
+}
+
+/// PEM/OpenSSH private key blocks (and PGP blocks, which are just as
+/// sensitive).
+fn looks_like_private_key(text: &str) -> bool {
+    text.contains("-----BEGIN") || text.contains("PRIVATE KEY")
+}
+
+/// A line that labels a credential directly: `password: hunter2`,
+/// `passwd = hunter2`, etc.
+fn looks_like_credential_line(text: &str) -> bool {
+    let first = text.lines().next().unwrap_or("").trim().to_ascii_lowercase();
+    const PREFIXES: [&str; 8] = [
+        "password:",
+        "password =",
+        "password=",
+        "passwd:",
+        "passwd =",
+        "passwd=",
+        "pw:",
+        "pass:",
+    ];
+    PREFIXES.iter().any(|p| first.starts_with(p))
+}
+
+/// A one-time code: a short copy that mentions a code keyword and contains
+/// a 6–8 digit run. A *bare* 6-digit number is deliberately not flagged —
+/// too many legitimate numbers have that shape.
+fn looks_like_otp(text: &str) -> bool {
+    if text.len() > 60 {
+        return false;
+    }
+    let lower = text.to_ascii_lowercase();
+    const LABELS: [&str; 8] = [
+        "code", "otp", "verification", "verif", "2fa", "passcode", "one-time", "one time",
+    ];
+    if !LABELS.iter().any(|k| lower.contains(k)) {
+        return false;
+    }
+    has_digit_run(text, 6, 8)
+}
+
+/// Credit card numbers: 13–19 digit runs (allowing spaces/dashes between
+/// groups) that pass the Luhn check.
+fn contains_card_number(text: &str) -> bool {
+    let mut run = String::new();
+    for c in text.chars() {
+        if c.is_ascii_digit() {
+            run.push(c);
+        } else if (c == ' ' || c == '-') && !run.is_empty() {
+            run.push(' '); // keep one run across group separators
+        } else {
+            if card_like(&run) {
+                return true;
+            }
+            run.clear();
+        }
+    }
+    card_like(&run)
+}
+
+fn card_like(run: &str) -> bool {
+    let digits: String = run.chars().filter(|c| c.is_ascii_digit()).collect();
+    (13..=19).contains(&digits.len()) && luhn_valid(&digits)
+}
+
+fn luhn_valid(digits: &str) -> bool {
+    let mut sum: u32 = 0;
+    let mut double = false;
+    for b in digits.bytes().rev() {
+        let mut d = (b - b'0') as u32;
+        if double {
+            d *= 2;
+            if d > 9 {
+                d -= 9;
+            }
+        }
+        sum += d;
+        double = !double;
+    }
+    sum.is_multiple_of(10)
+}
+
+/// Well-known API token prefixes (OpenAI, GitHub, Slack, AWS access keys).
+fn contains_api_token(text: &str) -> bool {
+    const PREFIXES: [&str; 7] = ["sk-", "sk-proj-", "ghp_", "gho_", "xoxb-", "xoxp-", "AKIA"];
+    PREFIXES.iter().any(|p| text.contains(p))
+}
+
+fn has_digit_run(text: &str, min: usize, max: usize) -> bool {
+    let mut run = 0usize;
+    for c in text.chars() {
+        if c.is_ascii_digit() {
+            run += 1;
+        } else {
+            if run >= min && run <= max {
+                return true;
+            }
+            run = 0;
+        }
+    }
+    run >= min && run <= max
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn private_key_blocks_are_sensitive() {
+        assert!(is_sensitive(
+            "-----BEGIN RSA PRIVATE KEY-----\nMIIEowIBAAKCAQEA..."
+        ));
+        assert!(is_sensitive("-----BEGIN PGP MESSAGE-----"));
+    }
+
+    #[test]
+    fn credential_labeled_lines_are_sensitive() {
+        assert!(is_sensitive("password: hunter2"));
+        assert!(is_sensitive("Password = correct horse battery staple"));
+        assert!(is_sensitive("passwd=hunter2"));
+    }
+
+    #[test]
+    fn labeled_one_time_codes_are_sensitive() {
+        assert!(is_sensitive("Your verification code is 483920"));
+        assert!(is_sensitive("483920 is your code"));
+        assert!(is_sensitive("OTP: 12345678"));
+    }
+
+    #[test]
+    fn bare_numbers_are_not_sensitive() {
+        // A bare 6-digit number is too ambiguous to drop on purpose.
+        assert!(!is_sensitive("483920"));
+        assert!(!is_sensitive("The answer is 42"));
+        // A long random number that happens to be 16 digits but fails Luhn.
+        assert!(!is_sensitive("1234567890123456"));
+    }
+
+    #[test]
+    fn luhn_valid_card_numbers_are_sensitive() {
+        assert!(is_sensitive("4111 1111 1111 1111"));
+        assert!(is_sensitive("card number: 4111-1111-1111-1111"));
+    }
+
+    #[test]
+    fn api_token_prefixes_are_sensitive() {
+        assert!(is_sensitive("sk-proj-abc123def456"));
+        assert!(is_sensitive("ghp_1234567890abcdefghijklmnopqrstuvwxyz"));
+        assert!(is_sensitive("AKIAIOSFODNN7EXAMPLE"));
+    }
+
+    #[test]
+    fn normal_prose_is_not_sensitive() {
+        assert!(!is_sensitive("just a normal sentence someone copied"));
+        assert!(!is_sensitive("here is the plan for next week"));
+        assert!(!is_sensitive("the password field in the form is empty"));
+    }
+}

breadclipd/src/main.rs

@@ -1,8 +1,11 @@
 mod content_kind;
+mod ignore_rules;
 
 use bread_utils::bread_client::BreadClient;
-use breadclip_core::HistoryDb;
-use std::{env, fs, path::PathBuf, process::Command, thread, time::Duration};
+use bread_utils::singleton::{try_acquire, Acquire};
+use breadclip_core::{CaptureSource, HistoryDb};
+use serde_json::Value;
+use std::{env, io::Read, process::Command, thread, time::Duration};
 
 /// This app's id in bread's sibling-app namespace registry
 /// (`bread_shared::apps::KNOWN_APPS`) — events are published as
@@ -14,15 +17,52 @@ const APP_ID: &str = "clip";
 /// needed for a single internal flag).
 const CAPTURE_FLAG: &str = "--capture-once";
 
+/// Second positional arg distinguishing a capture spawned by the primary-
+/// selection watcher from one spawned by the regular-clipboard watcher.
+const PRIMARY_FLAG: &str = "--primary";
+
 /// MIME type convention (originating with KDE's Klipper) that password
 /// managers such as KeePassXC and Bitwarden set on clipboard content they
-/// own, signaling "don't persist this". Any offer advertising it is skipped
-/// entirely.
+/// own, signaling "don't persist this". Only reachable on the fallback
+/// capture path — when invoked through `wl-paste --watch`, `CLIPBOARD_STATE`
+/// carries the same information from the same event (see `capture_once`).
 const PASSWORD_HINT_MIME: &str = "x-kde-passwordManagerHint";
 
-fn get_available_types() -> Vec<String> {
-    Command::new("wl-paste")
-        .args(["--list-types"])
+/// Build a `wl-paste` command with a hard deadline. Without the timeout, a
+/// stalled selection offer (source app died mid-transfer, compositor never
+/// completes the handoff) leaves `wl-paste` blocked forever reading a pipe
+/// that never closes — and a forever-blocked fallback capture would eat a
+/// file descriptor per event over a long session. `timeout -k` guarantees a
+/// SIGKILL if the initial SIGTERM doesn't land. Only used by the manual
+/// fallback path (`capture_via_wl_paste`); the `--watch` path reads the
+/// content from stdin and so has nothing that can hang.
+fn wl_paste_cmd(primary: bool) -> Command {
+    let mut cmd = Command::new("timeout");
+    cmd.args(["-k", "2", "5", "wl-paste"]);
+    if primary {
+        cmd.arg("--primary");
+    }
+    cmd
+}
+
+/// `wl-paste --watch` sets this in the spawned command's environment for
+/// every clipboard event (see wl-paste(1)): `data` (read the content from
+/// stdin), `nil` (empty clipboard), `clear` (explicitly cleared), or
+/// `sensitive` (a password manager flagged the selection). Unset or
+/// unrecognized means we were not invoked by `--watch` (e.g. a manual
+/// `--capture-once` run) and we fall back to querying wl-paste directly.
+const CLIPBOARD_STATE: &str = "CLIPBOARD_STATE";
+
+/// Open the history DB with the retention caps from the user's config. The
+/// `--capture-once` processes run per clipboard event, so they load the
+/// config fresh rather than inheriting anything from the long-lived daemon.
+fn open_db() -> Result<HistoryDb, breadclip_core::HistoryError> {
+    HistoryDb::open_with(breadclip_core::config::load().retention)
+}
+
+fn get_available_types(primary: bool) -> Vec<String> {
+    wl_paste_cmd(primary)
+        .arg("--list-types")
         .output()
         .ok()
         .filter(|o| o.status.success())
@@ -31,8 +71,16 @@ fn get_available_types() -> Vec<String> {
         .unwrap_or_default()
 }
 
-fn get_clipboard_text() -> Option<String> {
-    let output = Command::new("wl-paste")
+fn get_clipboard_bytes(mime: &str, primary: bool) -> Option<Vec<u8>> {
+    let output = wl_paste_cmd(primary).args(["--type", mime]).output().ok()?;
+    if !output.status.success() || output.stdout.is_empty() {
+        return None;
+    }
+    Some(output.stdout)
+}
+
+fn get_clipboard_text(primary: bool) -> Option<String> {
+    let output = wl_paste_cmd(primary)
         .args(["--no-newline", "--type", "text/plain"])
         .output()
         .ok()?;
@@ -44,48 +92,80 @@ fn get_clipboard_text() -> Option<String> {
         .filter(|s| !s.trim().is_empty())
 }
 
-fn get_clipboard_image() -> Option<Vec<u8>> {
-    let output = Command::new("wl-paste")
-        .args(["--type", "image/png"])
-        .output()
-        .ok()?;
-    if !output.status.success() || output.stdout.is_empty() {
-        return None;
+/// Identify an image payload by its magic bytes. `wl-paste --watch` hands us
+/// the content on stdin without saying which offered type it picked, so the
+/// bytes are the only reliable signal — and they're exact, because watch
+/// mode never appends a trailing newline.
+fn sniff_image_mime(bytes: &[u8]) -> Option<&'static str> {
+    if bytes.starts_with(b"\x89PNG\r\n\x1a\n") {
+        return Some("image/png");
     }
-    Some(output.stdout)
+    if bytes.starts_with(&[0xFF, 0xD8, 0xFF]) {
+        return Some("image/jpeg");
+    }
+    None
 }
 
-fn lock_file() -> PathBuf {
-    env::var("XDG_RUNTIME_DIR")
-        .map(PathBuf::from)
-        .unwrap_or_else(|_| PathBuf::from("/tmp"))
-        .join("breadclipd.lock")
+fn source(primary: bool) -> CaptureSource {
+    if primary {
+        CaptureSource::Primary
+    } else {
+        CaptureSource::Clipboard
+    }
 }
 
-// Returns false if another instance is already running.
-fn acquire_lock() -> bool {
-    let path = lock_file();
-    if let Ok(content) = fs::read_to_string(&path) {
-        if let Ok(pid) = content.trim().parse::<u32>() {
-            let alive = fs::read_to_string(format!("/proc/{}/comm", pid))
-                .map(|s| s.trim() == "breadclipd")
-                .unwrap_or(false);
-            if alive {
-                eprintln!("breadclipd: already running (pid {})", pid);
-                return false;
+fn store_text(text: &str, primary: bool) {
+    let db = match open_db() {
+        Ok(db) => db,
+        Err(e) => {
+            eprintln!("breadclipd: failed to open database: {e}");
+            return;
         }
+    };
+    match db.insert_text(text, source(primary)) {
+        Ok(()) => emit_copied(content_kind::detect(text), text.len()),
+        Err(e) => eprintln!("breadclipd: insert text: {e}"),
     }
 }
-    let _ = fs::write(&path, std::process::id().to_string());
-    true
-}
 
-/// Invoked once per clipboard-change event (as the handler command for
-/// `wl-paste --watch`). Reads whatever is on the clipboard right now, skips
-/// it entirely if a password manager flagged it as sensitive, and otherwise
-/// persists a text or image entry to the history DB.
-fn capture_once() {
-    let types = get_available_types();
+fn store_image(bytes: Vec<u8>, mime: &str, primary: bool) {
+    let db = match open_db() {
+        Ok(db) => db,
+        Err(e) => {
+            eprintln!("breadclipd: failed to open database: {e}");
+            return;
+        }
+    };
+    match db.insert_image(&bytes, mime, source(primary)) {
+        Ok(()) => emit_copied("image", bytes.len()),
+        Err(e) => eprintln!("breadclipd: insert image: {e}"),
+    }
+}
+
+/// Capture path for `wl-paste --watch` invocations: the clipboard content
+/// arrives on our stdin together with its `CLIPBOARD_STATE` in one event, so
+/// the sensitive check and the data read can't race each other (two separate
+/// `wl-paste` calls could straddle a clipboard change mid-capture).
+fn capture_from_stdin(primary: bool) {
+    let mut buf = Vec::new();
+    if std::io::stdin().read_to_end(&mut buf).is_err() || buf.is_empty() {
+        return;
+    }
+    if let Some(mime) = sniff_image_mime(&buf) {
+        store_image(buf, mime, primary);
+    } else if let Ok(text) = String::from_utf8(buf) {
+        if !text.trim().is_empty() && !ignore_rules::is_sensitive(&text) {
+            store_text(&text, primary);
+        }
+    }
+}
+
+/// Fallback capture path for manual `--capture-once` invocations (no
+/// `CLIPBOARD_STATE` was set by wl-paste): ask wl-paste directly. Same
+/// behavior as the stdin path, including requesting whichever image type is
+/// actually offered rather than assuming PNG.
+fn capture_via_wl_paste(primary: bool) {
+    let types = get_available_types(primary);
 
     if types.iter().any(|t| t == PASSWORD_HINT_MIME) {
         // Password manager (KeePassXC, Bitwarden, etc.) marked this copy as
@@ -97,31 +177,42 @@ fn capture_once() {
     let has_image = types.iter().any(|t| t == "image/png" || t == "image/jpeg");
     let has_text = types.iter().any(|t| t.starts_with("text/"));
 
-    let db = match HistoryDb::open() {
-        Ok(db) => db,
-        Err(e) => {
-            eprintln!("breadclipd: failed to open database: {e}");
-            return;
-        }
-    };
-
     if has_image && !has_text {
-        if let Some(bytes) = get_clipboard_image() {
-            match db.insert_image(&bytes) {
-                Ok(()) => emit_copied("image", bytes.len()),
-                Err(e) => eprintln!("breadclipd: insert image: {e}"),
-            }
+        let mime = if types.iter().any(|t| t == "image/png") {
+            "image/png"
+        } else {
+            "image/jpeg"
+        };
+        if let Some(bytes) = get_clipboard_bytes(mime, primary) {
+            store_image(bytes, mime, primary);
         }
     } else if has_text {
-        if let Some(text) = get_clipboard_text() {
-            match db.insert_text(&text) {
-                Ok(()) => emit_copied(content_kind::detect(&text), text.len()),
-                Err(e) => eprintln!("breadclipd: insert text: {e}"),
+        if let Some(text) = get_clipboard_text(primary) {
+            if !ignore_rules::is_sensitive(&text) {
+                store_text(&text, primary);
             }
         }
     }
 }
 
+/// Invoked once per clipboard-change event (as the handler command for
+/// `wl-paste --watch`). Stores a text or image entry to the history DB,
+/// unless the selection was flagged as sensitive — in which case nothing is
+/// ever persisted.
+fn capture_once(primary: bool) {
+    match env::var(CLIPBOARD_STATE).as_deref() {
+        // Password manager (KeePassXC, Bitwarden, ...) flagged this selection
+        // as sensitive via `wl-copy --sensitive` — never persist it.
+        Ok("sensitive") => {}
+        // Empty clipboard (nil) or an explicit clear — nothing to store.
+        Ok("nil") | Ok("clear") => {}
+        Ok("data") => capture_from_stdin(primary),
+        // Unset or unrecognized: not invoked by `--watch`. Fall back to the
+        // direct wl-paste path rather than blocking on whatever stdin is.
+        _ => capture_via_wl_paste(primary),
+    }
+}
+
 /// Publishes `bread.clip.copied` into the bread event fabric. Fire-and-forget
 /// and non-fatal by design (`BreadClient::emit` never blocks or errors this
 /// caller) — breadd being absent or not installed must never affect
@@ -134,19 +225,18 @@ fn emit_copied(kind: &str, len: usize) {
     );
 }
 
-/// Reacts to `bread.command.clip.*` verbs. Only `clear` maps to real,
-/// existing breadclip functionality today — `pin`/`select` would need a new
-/// "pinned" concept that doesn't exist anywhere in the history DB schema,
-/// which is a real product decision for breadclip itself (does it want
-/// pinning at all, and what would the GTK UI for it look like?), not
-/// something to fabricate as a side effect of wiring up the event bus. See
+/// Reacts to `bread.command.clip.*` verbs. Only `clear` and `pin` map to
+/// real, existing breadclip functionality today — `select` would need a
+/// "remote-activate a row" concept the popup doesn't expose over the bus,
+/// which is a real product decision for breadclip itself, not something to
+/// fabricate as a side effect of wiring up the event bus. See
 /// `breadclip/EVENTS.md` for the honest current status.
 ///
-/// Emits `bread.clip.<verb>.done`/`.failed` per the confirmation convention
-/// in bread's Documentation.md — a module that started this command via
-/// `bread.wait`/`bread.wait_any` can await the real outcome instead of
-/// assuming success the moment it publishes the command.
-fn handle_command(event_name: &str) {
+/// Emits `bread.clip.<verb>.done`/`bread.clip.pinned`/`.failed` per the
+/// confirmation convention in bread's Documentation.md — a module that
+/// started this command via `bread.wait`/`bread.wait_any` can await the real
+/// outcome instead of assuming success the moment it publishes the command.
+fn handle_command(event_name: &str, data: &Value) {
     let Some(verb) = event_name.strip_prefix("bread.command.clip.") else {
         return;
     };
@@ -164,22 +254,121 @@ fn handle_command(event_name: &str) {
                 );
             }
         },
+        "pin" => match data.get("id").and_then(Value::as_i64) {
+            Some(id) => {
+                let pin = data.get("pin").and_then(Value::as_bool).unwrap_or(true);
+                match HistoryDb::open().and_then(|db| db.set_pinned(id, pin)) {
+                    Ok(()) => {
+                        eprintln!(
+                            "breadclipd: {} entry {id} via bread.command.clip.pin",
+                            if pin { "pinned" } else { "unpinned" }
+                        );
+                        BreadClient::connect(APP_ID).emit(
+                            "bread.clip.pinned",
+                            serde_json::json!({ "id": id, "pinned": pin }),
+                        );
+                    }
+                    Err(e) => {
+                        eprintln!("breadclipd: bread.command.clip.pin failed: {e}");
+                        BreadClient::connect(APP_ID).emit(
+                            "bread.clip.pin.failed",
+                            serde_json::json!({ "error": e.to_string() }),
+                        );
+                    }
+                }
+            }
+            None => {
+                eprintln!("breadclipd: bread.command.clip.pin missing 'id'");
+                BreadClient::connect(APP_ID).emit(
+                    "bread.clip.pin.failed",
+                    serde_json::json!({ "error": "missing 'id'" }),
+                );
+            }
+        },
         other => {
             eprintln!("breadclipd: ignoring unrecognized command verb '{other}'");
         }
     }
 }
 
+/// Backoff for the `wl-paste --watch` restart loops: 2s → 4s → 8s → 16s →
+/// 32s, capped at 30s. A dead compositor keeps the daemon alive without
+/// hammering either wl-paste or the journal.
+fn restart_delay(consecutive_failures: u32) -> Duration {
+    Duration::from_secs((2u64 << consecutive_failures.saturating_sub(1).min(4)).min(30))
+}
+
+/// Log a watch-loop restart, but only on state changes — the first failure
+/// announces it, then every 8th attempt. Otherwise a persistent failure
+/// would spam the journal every two seconds forever.
+fn log_restart(what: &str, detail: &str, attempt: u32) {
+    if attempt == 1 || attempt.is_multiple_of(8) {
+        eprintln!("breadclipd: {what} ({detail}), restarting (attempt {attempt})");
+    }
+}
+
+/// Runs one `wl-paste --watch` watcher for the daemon's whole lifetime,
+/// restarting it with capped backoff whenever it exits (e.g. the compositor
+/// connection dropped). `primary` selects the middle-click primary selection
+/// watcher (`--watch --primary`) instead of the regular clipboard watcher.
+fn watch_loop(exe: std::path::PathBuf, primary: bool) {
+    let label = if primary {
+        "wl-paste --watch (primary)"
+    } else {
+        "wl-paste --watch"
+    };
+    let mut consecutive_failures: u32 = 0;
+    loop {
+        // `wl-paste --watch <cmd>` runs <cmd> once per selection-change
+        // event instead of polling — zero idle cost between changes, and no
+        // forking wl-paste repeatedly. Each event re-invokes this same
+        // binary with --capture-once to do a single read-and-store pass.
+        let mut cmd = Command::new("wl-paste");
+        cmd.arg("--watch");
+        if primary {
+            cmd.arg("--primary");
+        }
+        cmd.arg(&exe).arg(CAPTURE_FLAG);
+        if primary {
+            cmd.arg(PRIMARY_FLAG);
+        }
+        let status = cmd.status();
+
+        match status {
+            Ok(s) => {
+                consecutive_failures += 1;
+                log_restart(label, &format!("{s}"), consecutive_failures);
+            }
+            Err(e) => {
+                consecutive_failures += 1;
+                log_restart(label, &e.to_string(), consecutive_failures);
+            }
+        }
+        thread::sleep(restart_delay(consecutive_failures));
+    }
+}
+
 fn main() {
     let args: Vec<String> = env::args().collect();
     if args.get(1).map(String::as_str) == Some(CAPTURE_FLAG) {
-        capture_once();
+        capture_once(args.iter().any(|a| a == PRIMARY_FLAG));
         return;
     }
 
-    if !acquire_lock() {
+    // flock(2)-based singleton (bread_utils::singleton): lock ownership is
+    // kernel-atomic and released automatically the instant this process
+    // dies, so there's no stale-pid-file case to reason about.
+    let _singleton = match try_acquire("breadclipd") {
+        Ok(Acquire::Acquired(guard)) => Some(guard),
+        Ok(Acquire::HeldByOther(pid)) => {
+            eprintln!("breadclipd: already running (pid {:?})", pid);
             std::process::exit(1);
         }
+        Err(e) => {
+            eprintln!("breadclipd: single-instance lock unavailable ({e}); continuing without it");
+            None
+        }
+    };
 
     // Wait briefly for WAYLAND_DISPLAY — common when started early in the session
     let mut retries = 0;
@@ -189,14 +378,14 @@ fn main() {
     }
     if env::var("WAYLAND_DISPLAY").is_err() {
         eprintln!("breadclipd: WAYLAND_DISPLAY not set after waiting, exiting");
-        let _ = fs::remove_file(lock_file());
         std::process::exit(1);
     }
 
+    let cfg = breadclip_core::config::load();
+
     // Fail fast (before we start watching) if the DB can't be opened.
-    if let Err(e) = HistoryDb::open() {
+    if let Err(e) = HistoryDb::open_with(cfg.retention) {
         eprintln!("breadclipd: failed to open database: {e}");
-        let _ = fs::remove_file(lock_file());
         std::process::exit(1);
     }
 
@@ -204,7 +393,6 @@ fn main() {
         Ok(p) => p,
         Err(e) => {
             eprintln!("breadclipd: could not resolve own executable path: {e}");
-            let _ = fs::remove_file(lock_file());
             std::process::exit(1);
         }
     };
@@ -219,29 +407,22 @@ fn main() {
     // delivering commands until it reconnects.
     let command_client = BreadClient::connect(APP_ID);
     let _commands = command_client.subscribe("bread.command.clip.**", |event| {
-        handle_command(&event.event);
+        handle_command(&event.event, &event.data);
     });
 
-    // `wl-paste --watch <cmd>` runs <cmd> once per clipboard-change event
-    // instead of polling — zero idle cost between changes, and no more
-    // forking wl-paste 4-6 times a second. Each event re-invokes this same
-    // binary with --capture-once to do a single read-and-store pass.
-    loop {
-        let status = Command::new("wl-paste")
-            .args(["--watch"])
-            .arg(&exe)
-            .arg(CAPTURE_FLAG)
-            .status();
+    // Regular-clipboard watcher, always on.
+    let clipboard_exe = exe.clone();
+    thread::spawn(move || watch_loop(clipboard_exe, false));
+    // Primary-selection watcher (middle-click), opt-in via config.
+    if cfg.capture_primary {
+        eprintln!("breadclipd: watching primary selection");
+        thread::spawn(move || watch_loop(exe, true));
+    }
 
-        match status {
-            Ok(s) => {
-                eprintln!("breadclipd: wl-paste --watch exited ({s}), restarting in 2s");
-            }
-            Err(e) => {
-                eprintln!("breadclipd: failed to spawn wl-paste --watch: {e}, retrying in 2s");
-            }
-        }
-        thread::sleep(Duration::from_secs(2));
+    // The watcher threads own the daemon's lifetime; this thread just keeps
+    // the process (and the singleton guard above) alive.
+    loop {
+        thread::sleep(Duration::from_secs(3600));
     }
 }
 
@@ -267,6 +448,22 @@ mod tests {
         f();
     }
 
+    #[test]
+    fn sniff_image_mime_detects_png_and_jpeg() {
+        assert_eq!(sniff_image_mime(b"\x89PNG\r\n\x1a\nrest"), Some("image/png"));
+        assert_eq!(sniff_image_mime(&[0xFF, 0xD8, 0xFF, 0xE0]), Some("image/jpeg"));
+        assert_eq!(sniff_image_mime(b"plain text"), None);
+    }
+
+    #[test]
+    fn restart_delay_backs_off_and_caps() {
+        assert_eq!(restart_delay(1), Duration::from_secs(2));
+        assert_eq!(restart_delay(2), Duration::from_secs(4));
+        assert_eq!(restart_delay(3), Duration::from_secs(8));
+        assert_eq!(restart_delay(4), Duration::from_secs(16));
+        assert_eq!(restart_delay(100), Duration::from_secs(30));
+    }
+
     #[test]
     fn handle_command_clear_empties_history_even_with_no_daemon_reachable() {
         // The point of this test: handle_command's actual effect (clearing
@@ -276,23 +473,57 @@ mod tests {
         // made contingent on the emit succeeding, this test would fail.
         with_isolated_history(|| {
             let db = HistoryDb::open().expect("open history db");
-            db.insert_text("something to clear").unwrap();
+            db.insert_text("something to clear", CaptureSource::Clipboard).unwrap();
             assert_eq!(db.list_entries(10).unwrap().len(), 1);
 
-            handle_command("bread.command.clip.clear");
+            handle_command("bread.command.clip.clear", &serde_json::json!({}));
 
             let db = HistoryDb::open().expect("reopen history db");
             assert!(db.list_entries(10).unwrap().is_empty());
         });
     }
 
+    #[test]
+    fn handle_command_pin_toggles_pinned_state() {
+        with_isolated_history(|| {
+            let db = HistoryDb::open().expect("open history db");
+            db.insert_text("to pin", CaptureSource::Clipboard).unwrap();
+            let id = db.list_entries(10).unwrap()[0].id;
+            assert!(!db.list_entries(10).unwrap()[0].pinned);
+
+            handle_command(
+                "bread.command.clip.pin",
+                &serde_json::json!({ "id": id, "pin": true }),
+            );
+            assert!(db.list_entries(10).unwrap()[0].pinned);
+
+            handle_command(
+                "bread.command.clip.pin",
+                &serde_json::json!({ "id": id, "pin": false }),
+            );
+            assert!(!db.list_entries(10).unwrap()[0].pinned);
+        });
+    }
+
+    #[test]
+    fn handle_command_pin_without_id_is_a_no_op() {
+        with_isolated_history(|| {
+            let db = HistoryDb::open().expect("open history db");
+            db.insert_text("untouched", CaptureSource::Clipboard).unwrap();
+
+            handle_command("bread.command.clip.pin", &serde_json::json!({}));
+
+            assert!(!db.list_entries(10).unwrap()[0].pinned);
+        });
+    }
+
     #[test]
     fn handle_command_ignores_unrecognized_verb() {
         with_isolated_history(|| {
             let db = HistoryDb::open().expect("open history db");
-            db.insert_text("should survive").unwrap();
+            db.insert_text("should survive", CaptureSource::Clipboard).unwrap();
 
-            handle_command("bread.command.clip.pin");
+            handle_command("bread.command.clip.select", &serde_json::json!({}));
 
             let db = HistoryDb::open().expect("reopen history db");
             assert_eq!(
@@ -307,11 +538,11 @@ mod tests {
     fn handle_command_ignores_events_outside_its_own_command_namespace() {
         with_isolated_history(|| {
             let db = HistoryDb::open().expect("open history db");
-            db.insert_text("should survive").unwrap();
+            db.insert_text("should survive", CaptureSource::Clipboard).unwrap();
 
             // Not a `bread.command.clip.*` event at all — must be a no-op.
-            handle_command("bread.command.pad.clear");
-            handle_command("bread.clip.copied");
+            handle_command("bread.command.pad.clear", &serde_json::json!({}));
+            handle_command("bread.clip.copied", &serde_json::json!({}));
 
             let db = HistoryDb::open().expect("reopen history db");
             assert_eq!(db.list_entries(10).unwrap().len(), 1);

contrib/breadclipd.service

@@ -11,6 +11,11 @@ ExecStart=%h/.cargo/bin/breadclipd
 Restart=on-failure
 RestartSec=2
 
+# wl-paste/wl-copy live on the user's PATH (e.g. ~/.local/bin or
+# ~/.cargo/bin via bakery/cargo install), which systemd user services don't
+# inherit by default — without this, breadclipd starts but captures nothing.
+Environment=PATH=%h/.local/bin:%h/.cargo/bin:/usr/local/bin:/usr/bin:/bin
+
 # Forward stdout/stderr to the journal so `journalctl --user -u breadclipd` works
 StandardOutput=journal
 StandardError=journal